Model Router engine — route requests across existing model engines

src/prerna/engine/api/IModelRouterEngine.java

@@ -0,0 +1,79 @@
+/*******************************************************************************
+ * Copyright 2015 Defense Health Agency (DHA)
+ *
+ * If your use of this software does not include any GPLv2 components:
+ * 	Licensed under the Apache License, Version 2.0 (the "License");
+ * 	you may not use this file except in compliance with the License.
+ * 	You may obtain a copy of the License at
+ *
+ * 	  http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * 	Unless required by applicable law or agreed to in writing, software
+ * 	distributed under the License is distributed on an "AS IS" BASIS,
+ * 	WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * 	See the License for the specific language governing permissions and
+ * 	limitations under the License.
+ * ----------------------------------------------------------------------------
+ * If your use of this software includes any GPLv2 components:
+ * 	This program is free software; you can redistribute it and/or
+ * 	modify it under the terms of the GNU General Public License
+ * 	as published by the Free Software Foundation; either version 2
+ * 	of the License, or (at your option) any later version.
+ *
+ * 	This program is distributed in the hope that it will be useful,
+ * 	but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * 	MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * 	GNU General Public License for more details.
+ *******************************************************************************/
+package prerna.engine.api;
+
+import java.io.IOException;
+
+import prerna.logging.IgnoreEngineLogging;
+
+/**
+ * A model engine that delegates each ask to one of several backing model
+ * engines based on a routing configuration it can reload at runtime.
+ *
+ * <p>These methods are declared on an interface rather than only on the
+ * implementation because Utility.getModel returns a dynamic proxy over the
+ * engine's interfaces (see EngineProxyFactory) - the concrete engine is
+ * unreachable through it, so a cast to the implementation class fails. Callers
+ * that need the routing config work against this interface and check
+ * {@code instanceof IModelRouterEngine} instead of the implementation class.
+ *
+ * <p>All methods are marked {@link IgnoreEngineLogging}: they are admin-time
+ * configuration operations, not model calls, so they should neither produce
+ * engine audit rows nor be run through the guardrail pipelines.
+ */
+public interface IModelRouterEngine extends IModelEngine {
+
+	/**
+	 * Raw contents of the routing config file, for the settings UI.
+	 *
+	 * @return the config file contents
+	 * @throws IOException if the config file is missing or unreadable
+	 */
+	@IgnoreEngineLogging
+	String readConfigJson() throws IOException;
+
+	/**
+	 * Validates the given JSON, persists it to the config file, and applies it to
+	 * the live instance. Nothing is written when validation fails.
+	 *
+	 * @param json the new routing config
+	 * @throws IOException if the config file cannot be written
+	 */
+	@IgnoreEngineLogging
+	void updateConfig(String json) throws IOException;
+
+	/**
+	 * Re-reads and applies the config file on the live instance, picking up an
+	 * edit made outside of {@link #updateConfig(String)}.
+	 *
+	 * @throws IOException if the config file is missing or unreadable
+	 */
+	@IgnoreEngineLogging
+	void reloadConfig() throws IOException;
+
+}

src/prerna/engine/api/ModelTypeEnum.java

@@ -35,6 +35,7 @@
 import prerna.engine.impl.model.KServeImageEngine;
 import prerna.engine.impl.model.KServeTTSEngine;
 import prerna.engine.impl.model.KServeVisionEngine;
+import prerna.engine.impl.model.ModelRouterEngine;
 import prerna.engine.impl.model.NEREngine;
 import prerna.engine.impl.model.OpenAiEngine;
 import prerna.engine.impl.model.TextEmbeddingsEngine;
@@ -63,6 +64,9 @@ public enum ModelTypeEnum {
 	REMOTE("REMOTE", RemoteModelEngine.class.getName()),
 	TEXT_EMBEDDINGS("TEXT_EMBEDDINGS", TextEmbeddingsEngine.class.getName()),
 	TEXT_GENERATION("TEXT_GENERATION", TextGenerationEngine.class.getName()),
+
+	// routing engine - dispatches to backing engines per its assets/router.json config
+	MODEL_ROUTER("MODEL_ROUTER", ModelRouterEngine.class.getName()),
 	;
 	// @formatter:on
 

src/prerna/engine/impl/model/AbstractModelEngine.java

@@ -500,7 +500,7 @@ public AskModelEngineResponse askRoom(String question, Room room, AbstractMessag
 				Thread inferenceRecorder = new Thread(new ModelEngineInferenceLogsWorker (
 						/*messageId*/ inputMessage.getMessageId(),
 						/*transactionId*/askModelResponse.getMessageId(),
-						/*messageMethod*/"ask",
+						/*messageMethod*/inferenceLogMessageMethod("ask"),
 						/*engine*/this,
 						/*insightId*/room.getInsight().getInsightId(),
 						/*projectContextId*/room.getInsight().getContextProjectId(),
@@ -573,6 +573,16 @@ public AskModelEngineResponse askRoom(String question, Room room, AbstractMessag
 		}
 	}
 
+	/**
+	 * messageMethod recorded on inference log rows written by this engine.
+	 * Delegating engines (e.g. the model router) override this to tag their
+	 * rows, so ask-history queries and usage aggregations can separate the
+	 * delegating row from the actual model call.
+	 */
+	protected String inferenceLogMessageMethod(String method) {
+		return method;
+	}
+
 	@Override
 	@Deprecated
 	public AskModelEngineResponse ask(String question, String context, Insight insight,
@@ -613,7 +623,7 @@ public EmbeddingsModelEngineResponse embeddings(List<String> stringsToEmbed, Ins
 			Thread inferenceRecorder = new Thread(new ModelEngineInferenceLogsWorker (
 					/*messageId*/messageId,
 					/*transactionId*/messageId,
-					/*messageMethod*/"embeddings",
+					/*messageMethod*/inferenceLogMessageMethod("embeddings"),
 					/*engine*/this,
 					/*insightId*/insight.getInsightId(),
 					/*projectContextId*/insight.getContextProjectId(),